Re: <link> and <param> in SVG (SVG ACTION-2534)

Hi Doug,

On May 11, 2009, at 01:54 , Doug Schepers wrote:
> Robin Berjon wrote (on 5/5/09 8:21 AM):
>> I think that we're much better off with separating the semantics of  
>> href
>> and src, and not adding two attributes (which then need one to take
>> precedence, and complicates generic scripting, etc.).
>
> Hmmm... I'm not sure what you're saying here.  Could you restate it?

Simply that it ought to be <svg:a href=''> and <svg:script src=''>,  
not the addition of both attributes everywhere there was XLink, with  
no distinguishing semantics.

>> This is interesting, but may prove tricky. The inheritance of <use>
>> content is already expletived up and the cause of a number of
>> implementation annoyances (notably caching optimisations), throwing  
>> in
>> parameters might make it even more complex.
>
> Can you expand on what problems you've seen?

The major problem is that <use> has a wacky inheritance tree: the  
conceptual cloning copies over the properties from the CSS cascade,  
and then applies property inheritance from the starting point of  
reference. Implementation-wise that means that you have to special- 
case inheritance for use, and also that you can't optimise the  
rendering of the referenced element to reuse it whenever it is <use>d  
because each will have a different cascade context.

The value to end-users is minimal: you can't usefully, for any case  
more complex than a couple elements, do anything much to override the  
styling of the referenced element; at the same time the cost to  
implementers is high. What I'm saying is that rather than trying to  
fix it we shouldn't simply move on to phase it out and replace it.

-- 
Robin Berjon - http://berjon.com/
     Feel like hiring me? Go to http://robineko.com/

Received on Wednesday, 13 May 2009 10:30:16 UTC